Aims. Clusters of stars are key objects for studying the dynamical and chemical evolution of the Galaxy and its neighbors, and are the most important laboratories to test the theory of stellar evolution. In particular, chemical composition is obtained from different kinds of stars (hot main-sequence stars, cool main-sequence stars, horizontal-branch stars, RGB stars) using different methodologies. Our first aim is to apply these methodologies to the stars of the open cluster NGC 6475 and, by obtaining a census of the most important elements, we will be able to test their consistency. Our second aim is to study the evolution of the surface chemical abundances as a function of the evolutionary phase of a star. We finally want to establish more robust fundamental parameters for this cluster. Methods. We selected high S/N high resolution spectra of 7 stars of the open cluster NGC 6475 from the ESO database covering the T eff range 4500-10 000 K and of luminosity class V (dwarf) and III (giants). We determined the chemical abundances of several elements. For hot stars (T eff > 9000 K), we applied the Balmer line fitting method to obtain atmospheric parameters, while for cool stars (T eff < 6500 K), the abundance equilibrium of FeI/II lines. For the two groups of stars, the use of different line-lists was mandatory. LTE approximation together with NLTE correction for some elements (C, N, O, Na, Mg) were applied. The abundances of many elements were obtained by measuring of the equivalent width of spectral lines. For those ones having only blended lines (O, He), real spectra were compared to synthetic ones. Hyperfine structure was taken into account for V and Ba. Results. First of all, we showed that the two methodologies we used give abundances that agree within the errors. This implies that no appreciable relative systematic effects are present for the derived chemical content of cool and hot stars. On the other hand, giants stars show clear chemical peculiarities with respect to the dwarfs affecting light elements (up to Si) and maybe Ba. This can be explained as an evolutionary effect. Then, once we had a new estimation of the metallicity for the cluster ([Fe/H] = +0.03 ± 0.02, [α/Fe] = −0.06 ± 0.02), we fitted suitable isochrones to the CMD of the cluster to obtain the basic parameters (E(B − V) = 0.08 ± 0.02, (m − M) 0 = 7.65 ± 0.05, Age = 200 ± 50).